I think it may be more accurate to say perceptual description of luminance may be exponential (that humans report the difference from 1 to 1/2 as being the same as 1/2 to 1/4). But I think this is all irrelevant if you are trying to reproduce tones correctly (other than it says the error-model is relative error on tones). So I would say it takes a very labored reading to think of the article statement as correct (as light falls of 1/R^2 not exponentially). But that also makes the article's point: if you are painting what is there you are more immune to wrong assumptions.

Not necessarily. If you are making a painting, you don't have the ability to make some sections literally brighter (higher luminescence) because you don't really have control over the lighting conditions, and oil paints only give you so much ability to adjust reflectance. So what you do is subtly adjust things like tone, texture, detail or technique to 'trick' the human perceptual system into thinking the difference is there. These tricks are what imparts that life-like quality into the old master's works.

The term “luminance” refers to a physical quantity, the intensity of the light. It can be measured in candelas/square meter. The relevant technical terms for the perceptual attributes of a color are “brightness” (which is the perception of how bright something is on an absolute scale given current adaptation of the eye/surroundings), and “lightness” (which is the perception of how bright something is relatively, taking into account knowledge of the scene lighting).

Luminance (light intensity) is going to fall off with the square of distance.

Brightness/lightness do not have an exponential relationship with luminance. The curve is (very) roughly like a square root.

In any event, the article was clearly mistaken about this point. It’s a fairly trivial mistake though. Vanity Fair’s fact checkers probably didn’t bother to ask an optics / color expert about the statement.